Abstract
Widespread ectopic activation of the ELT-7 transcription factor is capable of forcing direct cell fate conversion in fully differentiated and post-mitotic tissues. The mechanisms underlying this potent reprogramming ability, however, remain poorly understood. To explore the role of ELT-7 in reprogramming and to search for additional factors that can enable or prevent transdifferentiation into intestine-like tissue, multiple genetic approaches were utilized, including quantification of transgenic intestine-specific reporter constructs, elt-7 deletion mutants, and targeted RNAi screening. I find that elt-2p::gfp and ifb-2::cfp
reporter expression follows a reproducible and tissue-specific pattern during reprogramming, and that END-3-driven reprogramming is successful in an elt-7(-) mutant. I identify three chromatin-associated factors, epc-1, smo-1, and pyp-1 that may be important for somatic gonad-to-intestine transorganogenesis. These results suggest that direct reprogramming to an intestine cell fate involves recapitulation of embryonic-like intestine developmental events.
In addition, the complete formation of a secondary intestine in worms without a functional elt-7 gene shows that ELT-7 activity is not necessary for transorganogenesis. The vast majority of factors tested in the RNAi screens had no post-embryonic RNAi phenotypes or effects on reprogramming efficiency. In contrast, the three chromatin-associated factors identified all have an effect on gonad development and their inhibition interferes with ELT-7 directed reprogramming. These factors may each function in the establishment of a
chromatin state that is permissive to reprogramming to an intestine fate, although the molecular mechanisms involved remain to be elucidated. These studies provide insight into the dynamics of in vivo reprogramming, affirm the robust capability of the endoderm gene
34
regulatory network to redirect cell fate, and highlight factors that may determine the cellular contexts that allow for reprogramming into intestine in C. elegans.
Introduction
Embryonic development of the C. elegans intestine involves a core pathway of seven transcription factors that are activated in a sequential feed-forward cascade (Fig 1 and see also Chapter 1) (Maduro, 2015; Maduro et al., 2007; Maduro et al., 2015; Sommermann et al., 2010). This regulatory logic enables rapid and robust specification of intestine cell fate (Maduro, 2009; Maduro, 2015; Maduro et al., 2005; Raj et al., 2010). The rest of the cell types in the embryo are similarly specified and early in development the embryo completes the multipotency-to-commitment transition (MCT), after which cell fates stabilize and remain fixed and resistant to alteration for the remainder of the worm’s life (Djabrayan et al., 2012; Erdelyi et al., 2017; Rothman and Jarriault, 2019; Spickard et al., 2018; Yuzyuk et al., 2009). Previous research has revealed a novel and remarkable exception to this norm;
multiple factors within the C. elegans endoderm developmental transcription factor cascade, namely END-3, ELT-2, and ELT-7, each individually possess the ability to force cell fate changes resulting in the complete restructuring of an entire organ within the living animal (Riddle et al., 2013; Riddle et al., 2016). It is the ELT-7 transcription factor, however, that induces cellular reprogramming with the greatest efficiency.
What makes this particular factor such a potent activator of intestine gene expression in an ectopic cellular context? If ELT-7 contains a unique reprogramming ability, the
interconnectivity of the endoderm gene regulatory network (GRN) could function to activate ELT-7 from various initially expressed nodes. Alternatively, the robustness of the network as a whole may drive intestine fate acquisition regardless of the initial starting point. In addition to ELT-7 activity, what are the molecular characteristics that define the particular cellular contexts in which reprogramming into intestine is permissible?
36
In order to address these questions, I used RNA interference and genetic mutants to perturb the endoderm GRN and screened for factors that modulate the reprogramming ability of ELT-7. I also quantified the ectopic expression of intestine tissue-specific reporter
constructs to characterize the dynamics and sequence of in vivo reprogramming events. I tested 338 genes encoding nuclear proteins known to act directly on DNA or chromatin and performed analysis of intestine reporter expression to identify subtle and quantitative
differences in tissue-specific reprogramming, as well as rare, major effects on ELT-7 directed transorganogenesis efficiency.
In this chapter I show that transdifferentiation of the pharynx and transorganogenesis of the somatic gonad into intestine both follow a sequence of events similar to embryonic intestine development, but with differing temporal dynamics, suggesting overlapping but partially distinct reprogramming mechanisms in these tissues. I also show that somatic gonad-to-intestine transorganogenesis can occur in worms without functional ELT-7, revealing that this factor is not necessary for reprogramming to an intestine fate. I further report the results of the candidate nuclear factor RNAi screening. Although the majority of gene knockdowns had no effect on ELT-7 directed reprogramming, I found three factors that appear to play a role in transorganogenesis: epc-1/E(Pc), smo-1/SUMO1, and pyp-1/PPA1.
These studies affirm the potency of ELT-7 and the intestine GRN to manipulate cell fates and provide insight into the mechanisms of in vivo cellular reprogramming and
transorganogenesis.
Results
Gut specific transgene markers suggest redeployment of gut development in pharynx and somatic gonad tissues
The dramatic changes in cell fate after widespread activation of the transcription factor ELT-7 in C. elegans have been confirmed by multiple lines of evidence when observing worms around 48 hours post activation (Riddle et al., 2013; Riddle et al., 2016). Most striking is electron microscopic observations revealing the reprogrammed uterus to be virtually indistinguishable from the endogenous intestine at the ultrastructural level (Riddle et al., 2016). While this endpoint of transorganogenesis is remarkable, the dynamics of the process are not well characterized. The rate and efficiency of reprogramming in the cells of the pharynx and somatic gonad could indicate how cells respond to ectopic ELT-7. Cells that are more robust in their GRNs and that resist reprogramming more or less strongly may have different dynamics from cells that are more susceptible to reprogramming. Furthermore, as there are different tissues that become reprogrammed, there could be shared or disparate mechanisms that enable them to be reprogrammed into intestine-like cells. Similar temporal dynamics might suggest a similar mechanism, while different dynamics might suggest different mechanisms.
The dynamics of cellular reprogramming were determined by quantifying the percent of worms in a population that expressed either elt-2p::gfp or ifb-2::cfp transgenic markers in a particular tissue over a range of time points. The elt-2p::gfp transgene is a transcriptional reporter expressed early in the E-cell lineage and is continually expressed into adulthood.
The elt-2 gene can be activated by ELT-7 (Sommermann et al., 2010), so the transgene also
38
serves as a marker for ELT-7 transcriptional activator activity and an early marker for potential conversion to gut cell fate. The ifb-2::cfp transgene is a translational reporter specifically expressed in the intestine from late embryonic development throughout adulthood (Bossinger et al., 2004; Hüsken et al., 2008). ifb-2 encodes an intermediate
filament protein which is integrated into the terminal web cytoskeleton of the intestine and is an indicator of differentiation into intestinal cell fate (Dodemont et al., 1994; Jahnel et al., 2016; Karabinos et al., 2001). The experiment was carried out by feeding multiple plates of synchronized hsp::elt-7 transgenic L1 larvae for approximately two days, until the early L4 stage, and then exposing the worms to a 330C heat shock for 30 minutes. For each time point a separate population of worms was scored for whether elt-2p::gfp and ifb-2::cfp was
prominently expressed in either the pharynx, somatic gonad, or other tissues (Fig 2).
After widespread activation of hsp::elt-7, elt-2p::gfp is rapidly expressed and by 6 hours GFP is visible in all major cell types, including hypodermis, muscles, and neurons. It is particularly strongly expressed in the pharynx (Fig 2A-D). elt-2p::gfp expression is more gradually observed in the somatic gonad and on occasion is absent in worms that later contain a well-formed ectopic lumen. The expression of ifb-2::cfp takes longer to be visibly expressed than elt-2p::gfp in all tissues. It takes 8-12 hours on average for ifb-2::cfp to be detectably expressed in the pharynx, and 12-16 hours in the somatic gonad (Fig 2G). Ectopic ifb-2::cfp is also expressed at very low levels in various tissues other than the intestine, pharynx, and somatic gonad (Fig 2A). Expression of “widespread” ifb-2::cfp increases modestly over time until the point when worms are very sick or necrotic. This widespread expression is in the form of small puncta, appearing individually or in small clusters scattered throughout the animal, often in the hypodermis. Owing to their faint and stochastic nature,
precise quantification of the size, intensity, or number of widespread ectopic ifb-2::cfp puncta has not been achieved.
In embryonic development, elt-2 is activated early in the endodermal blastomeres (Fukushige et al., 1998), and ifb-2 is a downstream target that is activated a few hours later during differentiation of the intestine (McGhee et al., 2009). The dynamics of elt-2p::gfp and ifb-2::cfp expression after widespread ELT-7 activation follows the same temporal order, which suggests that transdifferentiation of pharynx and/or somatic gonad cells into intestine-like cells may occur similarly to embryonic differentiation of the intestine. There is, however, a notable difference in the dynamics of elt-2p::gfp and ifb-2::cfp expression in the pharynx versus the somatic gonad. This raises the possibility that there is a different mechanism for each tissue that enables ELT-7-directed reprogramming. Other evidence exists to support this: for example, while PHA-4/FoxO activity during embryogenesis is necessary for later reprogramming, knockdown of pha-4 during larval somatic gonad development does not prevent reprogramming of the uterus and spermatheca (Riddle et al., 2016).
Widespread ectopic elt-2p::gfp expression in other tissues is transient compared to that in pharynx and somatic gonad but is still visible for over 48 hours in most worms. Although widespread ectopic ifb-2::cfp expression is relatively insignificant, indicating that these other tissues are not being fully reprogrammed, the prolonged duration of elt-2p::gfp could suggest that in a few other cell types ELT-7 continues to activate transcription of target genes.
Continued ELT-7 activity could significantly disrupt the transcriptional state of widespread tissues as they would need to actively repress intestine gene networks while maintaining their correct regulatory network. This idea is explored further in Chapter 3. Alternatively, GFP is a
40
fairly stable protein (Bokman and Ward, 1981; Chalfie, 1995), and a slow fading away of a briefly expressed reporter could be explained by a long half-life.
ELT-7 is sufficient, but not necessary, for cellular reprogramming
Widespread ectopic activation of a single transcription factor is capable of inducing somatic gonad-to-intestine transorganogenesis. The transcription factor that achieves this with the highest efficiency is ELT-7; however, both ELT-2 and END-3 are also able to induce transorganogenesis (Riddle et al., 2016). All of these proteins are GATA factors that are specifically expressed in the E-cell lineage to promote intestinal cell fate (Fig 1). ELT-7 and ELT-2 are both terminal differentiation factors that activate transcription of each other and themselves. END-3 acts more upstream and is transiently expressed during embryonic development, and it has also been observed to have the lowest reprogramming efficiency of the three GATA factors.
Does the ability to induce transorganogenesis depend on the ability to activate ELT-7 in a given cell type? The reprogramming pathway might conceivably converge on one particular transcription factor in the endoderm GRN. There could be a particular biochemical property of ELT-7 that enables it to initiate cell fate change; for example, the ELT-7 protein is smaller than ELT-2, the larger ELT-7 isoform contains 198 amino acids and has a molecular weight of 23.1 kD which may allow it to penetrate deeper into condensed chromatin in order to activate intestinal genes. ELT-2, on the other hand, contains 433 amino acids and has a molecular weight of 47.1 kD. If such is the case, reprogramming initiated by ectopic expression of ELT-2 or END-3 would critically depend on being able to activate the endogenous elt-7 gene in additional tissues.
To test whether 7 is necessary for successful transorganogenesis I crossed the elt-7(ok835) mutation into the hsp::end-3 strain. The ok835 mutation consists of a 498 bp deletion which affects the 3’ intron and coding exon of the elt-7 gene (C. elegans
Consortium, 2012). This deletion mutation is assumed to result in a loss of function, although elt-7 is not an essential gene and elt-7(-) worms appear wild type. For reasons that are not obvious, the growth rate of the hsp::end-3; elt-7(-) strain was slightly slower than the hsp::end-3 strain. Synchronized L4 worms from both the hsp::end-3 control strain and hsp::end-3; elt-7(-) strain were heat shocked and scored for reprogramming after 24 hours (Fig 3). In the hsp::end-3 strain, 76% (n = 50) of worms show some signs of somatic gonad-to-intestine reprogramming as determined by the presence and morphology of ectopic IFB-2::CFP. In the hsp::end-3; elt-7(-) strain, 57% (n = 37) of worms show signs of
reprogramming. Although the loss of elt-7 results in a decrease in overall hsp::end-3
reprogramming efficiency, at a p < 0.05 significance level, this difference is not statistically significant (Fisher’s exact test p-value = 0.067).
These results indicate that an elt-7(-) mutation does not prevent somatic
gonad-to-intestine transorganogenesis driven by ectopic end-3 expression. Therefore, although ELT-7 is sufficient to induce transorganogenesis, ELT-7 transcription factor activity is not necessary for this process to occur.
Targeted reverse genetic screening for factors that determine susceptibility to reprogramming by ELT-7
Post embryonic, differentiated cells generally have very stable fates throughout an organism’s life span – an essential feature for continued health and longevity. In order to
42
reprogram cells, either from one fate directly to another or from a differentiated to a
pluripotent state, manipulation of multiple factors is often required (Cahan et al., 2014; Firas and Polo, 2017; Gao et al., 2017; Morris et al., 2014). In the now classic example of
reprogramming mammalian fibroblasts into induced pluripotent stem cells (iPSCs), four transcription factors were required, and the process was still highly inefficient (Takahashi and Yamanaka, 2006). Numerous subsequent studies that improved iPSC generation efficiency involved manipulation of additional factors, often inhibiting their function (Takahashi and Yamanaka, 2016; Vierbuchen and Wernig, 2012).
Only in rare instances can the activity of a single factor cause the conversion from one cell fate to another (Chambers and Studer, 2011; Chanda et al., 2014; Halder et al., 1995;
Weintraub et al., 1989). And yet, in C. elegans, activation of the elt-7 gene alone is capable of reprogramming the pharynx and somatic gonad into intestine-like tissue (Fig 4). On the other hand, the ectopic expression of elt-7 in all cells of the worm appears to be insufficient for reprogramming tissues other than the pharynx and somatic gonad. What, then, are the molecular characteristics that can define a cellular context that is permissible to
reprogramming to an intestine fate?
I performed a series of RNAi screens to identify candidate factors that are involved in determining susceptibility to ELT-7 directed reprogramming. The screens were designed to simultaneously test two complementary hypotheses. First, I hypothesized that there is a particular gene that enables ELT-7 to reprogram the pharynx and/or somatic gonad, the knockdown of which would result in a decrease in the efficiency of reprogramming. Second, I hypothesized that there is a gene that is expressed in certain tissues that functions to prevent reprogramming by ELT-7. Knockdown of such a gene would result in the reprogramming of
one or more additional tissue types. In theory, nearly any gene in the genome could play a role in reprogramming susceptibility. However, maintenance and manipulation of cell fate must depend on the cell’s overall transcriptional state. Therefore, factors acting within the nucleus, and especially on chromatin and DNA directly, were determined to be the top candidates for RNAi screening. Two separate methods were used to generate gene lists for RNAi screening and are described below. Furthermore, owing to the labor-intensive requirements for scoring reprogramming phenotypes, it was feasible to screen only ~2% of the genome, or about 350 genes using this approach.
Transcription factor binding site (TFBS) enrichment informed RNAi screening The primary objective of this screen was to identify genes that enable somatic gonad-to-gut transorganogenesis. A pilot transcriptional profiling experiment using mRNA-seq detected thousands of transcripts that are differentially expressed 3 hours after ectopic elt-7 expression [see Chapter 3 for discussion of the full transcriptional profiling study]. These early gene expression changes could be attributable to direct transcription factor activity of ELT-7 but may also be driven by increased activity of another, unknown transcription factor, that is promoting cellular reprogramming. In order to identify candidate transcription factors that contribute to early gene expression changes in worms undergoing transorganogenesis, I used the web-based tool oPOSSUM (http://opossum.cisreg.ca/oPOSSUM3/), which enables the detection of over-represented conserved TFBS in sets of genes (Kwon et al., 2012; Sui et al., 2005; Sui et al., 2007). Multiple oPOSSUM analyses were performed using the 2,000 most significantly upregulated transcripts at 3 hours post heat shock (hrs PHS) and TFBS enrichment scores were averaged to determine the transcription factor families with the most prevalent binding sites (Fig 5). Binding sites for GATA factors were the most significantly
44
enriched, followed by Forkhead, Homeobox, and High Mobility Group (HMG) transcription factors. To populate a candidate list of transcription factors for RNAi screening, most
members of the top families, as well as C. elegans homologs of the specific transcription factors with the most significant enrichment scores in the oPOSSUM database, were included. For example, all GATA factors in the C. elegans genome were screened, plus factors such as the vertebrate beta-beta-alpha-zinc finger transcription factor Evi1 homolog egl-43. Additional candidate factors of interest, such as the C. elegans chromodomain-containing gene cec-4, were included for a total of 118 genes tested in this screen; the L4440 empty RNAi vector was used as the control (Fig 6). Specific results are described in
following sections.
Chromatin-related factor RNAi screening
Forced cell fate reprogramming likely involves extensive chromatin remodeling and restructuring of DNA topology to allow for the required dramatic changes in gene expression (Beagan et al., 2016; Guo and Morris, 2017; Onder et al., 2012). Chromatin modifying factors that are specifically expressed in the pharynx or somatic gonad, either endogenously or in response to ectopic ELT-7, could be promoting reprogramming. For instance, a histone acetyl transferase complex that is active in the cells of the uterus may help to establish a chromatin state that is more permissible to ELT-7 binding, the knockdown of which would result in decreased transorganogenesis efficiency. Alternatively, chromatin-related factors expressed in other tissues might establish epigenetic barriers to ELT-7 directed
reprogramming, such as H3K9 methyltransferases, which can define transcriptionally silent heterochromatin (Ahringer and Gasser, 2018). Inhibition of such factors might allow for additional tissues to be reprogrammed by ELT-7 into intestine.
A previous screen for chromatin-related epigenetic barriers to cellular reprogramming of germ cells into neuronal cells in C. elegans successfully identified the histone chaperone LIN-53/RBBP4 (Tursun et al., 2011). To generate a candidate list of chromatin related genes for RNAi screening I combined lists of C. elegans chromatin factors from this previous study and other sources (Cui and Han, 2007; Tursun et al., 2011; Zuryn et al., 2014), and selected 220 genes to be targeted by RNAi (Fig 7).
The majority of gene targets showed no discernable developmental or reprogramming phenotypes when knocked down by RNAi. Knockdowns that had an effect on
reprogramming efficiency typically also had more severe developmental phenotypes such as a protruding vulva or sterility. In ELT-7 reprogrammed, L4440 RNAi controls, 89.4% (n = 772) of worms were scored as having an ectopic lumen (“lumen”). The most significant hits were considered to be conditions where ≤ 60% of worms showed the “lumen”
reprogramming phenotype, resulting in 19 potential hits. It should be noted, however, that the
reprogramming phenotype, resulting in 19 potential hits. It should be noted, however, that the